4 ms·
> Since it is a big wall of text, let us ask the subject to summarize. Let's see if it did a good job. That's the whole problem. Now you have to read the summa
by kranner 2y ago
> Since it is a big wall of text, let us ask the subject to summarize. Let's see if it did a good job.
That's the whole problem. Now you have to read the summary and the original, just to verify whether the summary was correct — especially if it's a niche topic (admittedly I'm going by the summary here).
- swyx 2y agoknown flawed drafts are useful even if known flawed :) we just need some way to quantify the error rate and gradually bring them down
- kranner 2y agoMaybe for a certain class of low-interest work, i.e. for texts where I don't care that I would miss something interesting. For things I care about, the error rate has to come down to zero or the possibly-flawed draft is really of no use.